Influence of Maternal Toxicity in Studies on Developmental Toxicity 2 March 2004, Berlin. Workshop Report No. 4

Transcription

1 Influence of Maternal Toxicity in Studies on Developmental Toxicity 2 March 2004, Berlin Workshop Report No. 4 Brussels, October 2004

2 ECETOC WORKSHOP REPORT No. 4 Copyright - ECETOC AISBL European Centre for Ecotoxicology and Toxicology of Chemicals 4 Avenue E. Van Nieuwenhuyse (Bte 6), B-1160 Brussels, Belgium. All rights reserved. No part of this publication may be reproduced, copied, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the copyright holder. Applications to reproduce, store, copy or translate should be made to the Secretary General. ECETOC welcomes such applications. Reference to the document, its title and summary may be copied or abstracted in data retrieval systems without subsequent reference. The content of this document has been prepared and reviewed by experts on behalf of ECETOC with all possible care and from the available scientific information. It is provided for information only. ECETOC cannot accept any responsibility or liability and does not provide a warranty for any use or interpretation of the material contained in the publication. The Workshop was sponsored by the Cefic Long-range Research Initiative (LRI) The programme was endorsed by the European Teratology Society

3 Influence of Maternal Toxicity in Studies on Developmental Toxicity CONTENTS 1. EXECUTIVE SUMMARY 1 2. WORKSHOP OVERVIEW 2 3. KEY DISCUSSION POINTS ON PLENARY LECTURES AND BREAKOUT GROUPS What should be the appropriate definition of maternal toxicity? Does pregnancy preclude the use of toxicity data from other studies? What is the impact of mode and route of chemical administration on the manifestations 11 of maternal and development toxicity? 3.4 Do further end points improve differentiation between maternal toxicity and 12 manifestations of developmental toxcity? 4. CONCLUSIONS AND RECOMMENDATIONS 14 APPENDIX 1: WORKSHOP PROGRAMME 16 APPENDIX 2: PRESENTATION ABSTRACTS 17 APPENDIX 3: BACKGROUND PAPER ON THE STATE OF THE SCIENCE 24 APPENDIX 4: LIST OF PARTICIPANTS 46 APPENDIX 5: ORGANISING COMMITTEE 48

4 1. EXECUTIVE SUMMARY The Workshop was convened to explore whether the criteria to assess the influence of toxic effects induced in the maternal organism on the development of the embryo/foetus could be improved for classification purposes. Several conclusions and recommendations were made: 1. Regulatory classification of chemicals that have caused signs of developmental toxicity relies on a hazard-based approach. It does not take full account of exposure conditions during normal handling and use. 2. The currently practiced European classification of chemicals that have shown evidence of developmental toxicity does not provide a clear distinction between developmental effects that occur directly on the embryo/foetus (primary effect) and those that are associated with chemically-induced perturbations of maternal homeostasis (secondary effect). The appropriate consideration of such effects has profound ramifications for the classification of chemicals. 3. Relevance of route and mode of administration of test chemicals, doseresponse data as well as toxicokinetics and toxicodynamics were recognised as important criteria that should be considered in the study design. The existing data are compelling, and one can no longer ignore the scientific evidence that the biological and toxicological effects differ profoundly depending on how the chemical enters into the body. 4. The use of expanded end points that are more sensitive to detect chemically induced maternal toxicity than those specified in the OECD 414 test guideline received endorsement from a majority of the participants. However, reservations from representatives of regulatory agencies meant that no true consensus could be reached. An experimental way forward to better define maternal and developmental toxicity in OECD 414 studies received broad support. New data with refined end points to detect the onset of maternal toxicity might improve the interpretation of the results of such studies and might provide a scientific basis for industry and regulators when making decisions leading to classification. 5. The majority of participants agreed with the recommendation to make the regulatory process reflect anticipated human exposure. This would mean that risk characterisation, as well as hazard identification, would become an important component in the classification of chemicals. 1

5 2. WORKSHOP OVERVIEW Background Maternal toxicity often occurs as a consequence of high doses required in developmental toxicity studies conducted according to current testing guidelines. Such maternal toxicity is likely to produce indirect effects on the progeny, for example through compromised placental circulation, which can be as consequential as effects detected from primary developmental toxicants. However, the effects of maternal toxicity need to be recognised as such, and distinguished from primary developmental toxicity. The reason for holding this Workshop was a response to a Cefic s Long-range Research Initiative (LRI) members survey. The LRI commissioned ECETOC to organise and report on the Workshop. All invited participants were provided with a Background Paper (attached) in advance of the Workshop meeting. The purpose of that document was to give an overview of the state of the science and provide the rationale for focusing the meeting on the OECD 414 developmental toxicity guideline-compliant study in rats. Objective of the Workshop The general objective of the Workshop was to build the grounds for finding consensus on the criteria for an appropriate interpretation of developmental effects occurring in the presence of maternal toxicity. The aim was to put strong emphasis on defining the actions needed to develop a better insight into the genuine incidence of such secondary effects. This new knowledge would be included in future study design, so that phenomena of maternal toxicity could be identified and distinguished from primary developmental toxicity, and would thereby lead to a more reliable hazard identification with impact on classification and labelling. Workshop Structure The Workshop was an interactive limited scale event with 42 invited participants, representing a balanced participation from regulatory agencies, academia and industry. The morning session of the Workshop started with four individual presentations (abstracts attached). 2

6 In the afternoon, Workshop participants separated into three breakout groups to address in more depth three topics (see specifics below) that built on the preceding individual presentations of the morning session. Plenary Lectures The opening presentation by Dr. Welsch reviewed the present OECD 414 test guidelines and their recorded end points. Equivocal outcomes of developmental toxicity studies carried out according to OECD 414 at times cause differences in interpretation by study directors and regulatory agencies. In those cases, the testing observations do not allow an unequivocal decision as to whether developmental toxicity encountered at doses that induced maternal toxicity was the result of direct effects of the test article on the embryo/foetus (a primary effect) or may have been caused indirectly by toxic effects on the maternal organism (a secondary effect). The colloquial term chemicals of the grey zone was introduced and used repeatedly during subsequent Workshop discussions. In particular that designation was applied in the context of the deliberations concerning expansion of the end points of maternal toxicity assessments with a small number of chemicals for which no unequivocal decision could be made as to a primary or secondary effect causing developmental toxicity. In the second presentation, Dr. Sullivan provided a historical perspective as to how the classification of chemicals for toxicity to reproduction was introduced in 1992 and how since then it has worked in practice. In the intervening more than 10 years of regulatory agency application of the Dangerous Substances Directive 69/549/EEC, it has become apparent that a hazard-based approach used in the interpretation of animal testing data has been strictly applied. This has caused classification and labelling of too many chemicals into categories that have had serious economic consequences for the manufacturers and no benefits to public health. There are now several initiatives underway in various expert working groups of the EU how to improve the interpretation of animal study data and the resulting regulatory approach. Dr. Sullivan concluded on the recommendation that a weight of evidence approach is necessary so that more attention is given to risk characterisation. The hazard identification-based classification and labelling that relies on the developmental toxicity outcome of effects elicited by any dose by any route should be reconsidered and requires revision. No account is currently taken of normal handling and use and the relevance of animal experimental data to humans. Exposure considerations of humans in the real world are thus missing in the regulatory process leading in some cases to unwarranted downstream consequences. 3

7 In the third presentation, Dr. Stahlmann illustrated with specific examples that high oral bolus doses, as they are generally applied in OECD 414 studies, may only be incompletely absorbed from the gastrointestinal tract of small laboratory animals. Thus, the systemic uptake and distribution (internal exposure) becomes much lower than anticipated. But pharmacokinetic/toxicokinetic measurements are typically not performed on agrochemicals and industrial chemicals during the entire process of developmental toxicity hazard evaluations. Studies with numerous substances have provided general insights about species and sex differences in chemical metabolism. Occasionally some data have been generated regarding metabolic disposition changes brought about by pregnancy. However, it is difficult to make generalisations, and a chemical-by-chemical approach seems to be required. Furthermore, metabolism of chemicals in the liver, during first pass following gastrointestinal tract absorption, can be highly dose-dependent and profoundly affect toxicokinetics. The metabolic fate thus becomes entirely different from that following another route or mode of administration, which would much more closely mimic human exposure during normal handling and use. The presentation concluded that major efforts should be made to generate pharmacokinetic/toxicokinetic data in developmental toxicity hazard identification studies. The route and mode of administration of the test chemical should reflect the anticipated human exposure as closely as possible, so that the data would provide a basis for rational comparisons between species. The fourth presentation by Dr. Daston addressed specifically maternal toxicity and concurrent developmental toxicity manifestations in rats. He described experimental results that have identified several potential mechanisms for maternally mediated developmental toxicity. The hypothesis has evolved that it is technically feasible to include markers for some common mechanisms into the standard testing protocols. Those factors go beyond the present simple end points of maternal feed consumption, body weight development and obvious manifestations of chemically induced toxicity (so called clinical signs ). The proposal made by Dr. Daston was to incorporate new biomarkers concerning the acute phase response (e.g. acute phase proteins in serum, serum zinc concentration, or hepatic metallothionein concentration) and potentially make those components an integral part of the study design if they turn out to be meaningful once explored with more compounds. The new information that would be obtained could make a useful contribution to the unsolved question as to what level of a maternal effect is needed to concomitantly cause developmental toxicity. If expanded end points are measured as part of dose-range finding studies, these biomarkers could serve to identify a maximum tolerable dose that triggers the onset of objectively measurable signs of toxicity for a subsequent definitive developmental toxicity hazard identification study. 4

8 These four presentations set the stage for a plenary discussion on the topic What should be the appropriate definition of maternal toxicity? This was first discussed during the morning session, and the debate was resumed later in the day as the closing agenda item. Breakout Group Reports 1. Does pregnancy preclude the use of toxicity data from other studies? 2. What is the impact of mode and route of chemical administration on the manifestations of maternal and developmental toxicity? 3. Do further end points improve differentiation between maternal toxicity and manifestations of developmental toxicity? Each participant had a free choice in selecting his/her topic for the breakout groups. This led to a roughly even distribution between the three topics. The outcome of these small group deliberations was then summarised by a group rapporteur in front of the plenary session for further discussion. Report on Topic #1: Does pregnancy preclude the use of toxicity data from other studies? Chairman: Dr. Sullivan, Rapporteur: Dr. Christian The report made to the plenary session stated that a decision could only be made on a case-by-case, i.e. chemical-by-chemical, basis. At the outset the feeling may be that observations from other studies would not be applicable to the design of a pending developmental toxicity study, but many other factors should be considered. The route and mode of administration in 28 and 90-day studies are often different (e.g. dosed feed vs. gavage) and do not involve pregnant females. Therefore, it is recognised, based on established toxicokinetic and toxicodynamic grounds, reviewed in Dr. Stahlmann s presentation, that such preceding studies might provide different results from toxicity observations in pregnant animals (see also Topic 2 breakout group report, below). Furthermore, the age of the test animals is often different, yet test species and strain may be common factors. For example, there may be qualitative similarities in clinical signs between nonpregnant and pregnant animals; however, it may be difficult to demonstrate quantitative equivalence. Nevertheless, the findings from 28- and 90-day studies may give important clues for consideration of potential target tissues. That information can be included in a subsequent dose-range-finding study in pregnant animals or risk considerations. The outcome of developmental toxicity dose range finding studies with daily body weight and feed consumption monitoring, including clinical observations and signs of toxicity, should then be placed into the context of other data and supplement the information of a full study. 5

9 The observations are appropriate for use in the identification of perturbed maternal homeostasis. Another point made by this breakout group was that the currently applied approach of EU regulatory agencies, based on hazard identification from developmental toxicity studies, is not taking into consideration exposure and human risk assessment. Report on Topic #2: What is the impact of mode and route of chemical administration on the manifestations of maternal and developmental toxicity? Chairman: Dr. Stahlmann, Rapporteur : Dr. Reuter This group reconfirmed what the database shows so clearly, i.e. OECD 414 testing is usually done by gavage. The chemical industry has tended to use this conservative oral route for testing that is endorsed in the revised OECD 414 test guideline of However, for classification and risk assessment the relevant routes of human exposure should be taken into consideration. Among them is entry of chemicals into the mammalian body by inhalation, via the dermal route and by more protracted oral intake with feed and drink than gavage provides. Returning to the conclusions arrived at with Topic #1, this group also emphasised that the situation is different for each chemical. Although desirable, for cost reasons comparative studies with different routes and modes of application in most cases are not feasible. The group endorsed the message, delivered by Dr. Stahlmann s presentation, that kinetic data are needed to compare the findings of different studies with variable modes of application, e.g. dermal study toxicokinetic data. Data should be available on the areas under the curve in time vs. concentration plots (= AUC), and skin penetration and metabolism would give a good assessment. This breakout group largely identified with the concepts presented earlier in the Workshop. The report noted the fact that at the present time there is insufficient information available about the impact of route and mode of application on manifestations of maternal toxicity. Data from a conventional OECD 414 study about a given chemical s effect on clinical chemistry and haematology should be established and compared with those from other conventional studies. Substance-specific effects and information from other studies should be used (e.g. haematology) to identify the most relevant end points and to supplement routine guideline parameters. 6

10 The group also commented specifically on some routes of administration that one might contemplate in the design of developmental hazard identification studies. They felt that on occasion subcutaneous injection might be appropriate. One can envision that this mode of administration could be useful if, for example, dermal exposure was to be anticipated in humans during normal use and handling. However, the test chemical could be a skin irritant on intact skin of rats, but was tolerated upon daily repeated depositions into the subcutaneous space. The advantage of subcutaneous injection is that it eliminates first pass liver metabolism and may lead to more complete absorption, while the disadvantage is extrapolation to the human routes of exposure. In contrast the group felt that intraperitoneal injections were inappropriate for reproductive and developmental toxicity studies. One additional topic addressed was the need for testing in rabbits as the second species. A clear preference was voiced for good toxicokinetic measurements in the rat, for mechanistic data and information from humans over screening data from a second animal species. Report on Topic #3: Do further end points improve differentiation between maternal toxicity and manifestations of developmental toxicity? Chairman: Dr. Daston, Rapporteurs: Dr. Hellwig and Dr. Holzum The participants of this breakout group reached consensus that added end points could help with a more precise differentiation in toxicity manifestations. There remained disagreement, however, as to whether added maternal response end points, in and of themselves, would be sufficient to alter the regulatory process for classifying agents as reproductive toxicants, assuming at present that the hazard based approach will be retained in the future. In particular, the group members based in regulatory agencies and involved in classification would like to see a high standard of proof that would include research data demonstrating the causal relationship between the added maternal end points and adverse developmental outcome. Furthermore, they would prefer to be convinced by data that there existed a (quantitative) relationship between the expanded maternal end points and adverse development. The critical question is whether the maternal response must reach a certain level of severity before it becomes developmentally adverse. Based on the presentations in the workshop, as well as research conducted in the laboratories of those present, it was felt that the new maternal end points would most likely add value to developmental toxicity studies. Among the expanded measurements would be those that evaluate the acute phase response (e.g. acute phase proteins in serum, serum zinc concentration, or hepatic metallothionein 7

11 concentration), haematology, clinical chemistry/clinical pathology, organ weights or histopathology of selected organs. Histopathology would be limited to those organs identified as targets in 28- or 90-day sub chronic studies. The first suggestion of this breakout group was to evaluate 5 or 6 compounds from among chemicals already tested in an OECD 414 study. These would be compounds that have caused developmental toxicity of equivocal origin (primary vs. secondary effects). Those chemicals from the grey zone should be examined for their acute phase response on the selected end points mentioned above, as a means of demonstrating the utility of those added criteria. Given the concerns of some participants in the breakout group, it would be important to supplement that information with experiments like those described in Dr. Daston s presentation in the acute phase response that is determined within hours after administration of the chemical. The new data might demonstrate the causal relationship between these end points of maternal toxicity and their association with abnormal development. Included should be a temporal correlation between the onset of maternal effects and the developmental effects, and dose-response measurements as regards the onset, magnitude and progression in intensity of maternal effects with temporal correlation and detection of an emerging threshold for abnormal development. It is conceivable that the dose-response data from studies with expanded and more sensitive end points could in some instances lead to a narrower dosing range in future developmental toxicity studies. It is well known that dose-response curves in pharmacology and toxicology can be very steep. A 3-4 fold increase in dose between treatment groups, as is often applied in developmental toxicology, may have profound impact and may be too high if new end points are applied to characterise the perturbation of maternal homeostasis. There are very few good dose response data available on the correlation between maternal and developmental toxicity. Without more sensitive end points the question cannot be decided at present. The second proposal made was to ask chemical companies to re-examine their toxicology databases, including developmental toxicity studies, for correlations between expanded maternal end points (e.g. clinical chemistry) and adverse developmental outcome since it is possible that such data have previously been collected. One concern was that measurements of the maternal end points conducted near term (rat gestation day 20) in the past may not be useful for those studies in which under the old test guidelines dosing ceased on gestation day 15. Some comparisons have already been done and concluded that maternal clinical chemistry effects were measurable on gestation day 20 when dosing had continued through gestation day 19; however, when dosing had ended on gestation day 15 8

12 those effects were not present. Therefore, only studies conducted with the new test guidelines may be meaningful for this retrospective analysis. A third suggestion was to hold a second workshop to evaluate actual data sets in which expanded maternal end points had been collected. This proposal would seem reasonable if the first recommendation to evaluate 5 or 6 chemicals from the grey zone would be acted upon. 9

13 3. KEY DISCUSSION POINTS ON PLENARY LECTURES AND BREAKOUT GROUPS 3.1 What should be the appropriate definition of maternal toxicity? In the final part of the Workshop the discussion of that topic was resumed in plenary session. As a take-off point for the debate the proposed definition of maternal toxicity was: It involves disturbance of maternal homeostasis that may affect normal development of the embryo/foetus. The latter effect is secondary to maternal toxicity without direct effects on the embryo/foetus. At the end of the discussions, this definition, without more elaborate qualifications, was not satisfactory for all participants of the Workshop. The tools to differentiate between primary and secondary effects include several general parameters that have been recorded all along in guideline-compliant studies, among them body weight gain, feed consumption and clinical observations for signs of toxicity. Among specific parameters that should be considered were: 1. Chemical-specific mechanistic information from other studies. 2. Additional measurements to evaluate maternal homeostasis, such as blood gases, haematology and key enzymatic parameters (e.g. liver enzyme induction). One argument made was that adverse effects on the embryo/foetus are the only relevant outcome. Under the present hazard-based regulatory system of classification it does not matter whether this is a primary or secondary effect or a combination of the two. The test guidelines encourage use of high dose studies, which depending on the toxic properties of the specific chemical may cause clear signs of parental/maternal toxicity. Even some maternal deaths are an acceptable study outcome. Changes in maternal homeostasis could be detected before overt toxicity is observed if more sensitive end points would be applied. That would meet the criterion of raising doses to maternally toxic levels. The extrapolation of animal developmental hazard evaluation data should be made in light of potential human exposures. Only risk-based decisions would be meaningful to protect people because that is the purpose of the entire testing program. 10

14 3.2 Does pregnancy preclude the use of toxicity data from other studies? Decisions can only be made on a chemical-by-chemical basis and many other factors should be considered. Despite differences in route and mode of administration in 28 and 90-day studies and lack of information from pregnant females, the findings from 28- and 90-day studies may give important clues for consideration of potential target tissues. The discussions also questioned the currently applied approach of EU regulatory agencies to classification of developmentally toxic chemicals based on hazard rather than on risk. 3.3 What is the impact of mode and route of chemical administration on the manifestations of maternal and developmental toxicity? The profound impact that mode and route of administration can have on toxicity was explicitly recognised. The route and mode of application chosen for animal hazard identification testing should mimic anticipated human exposure as closely as possible if the physical-chemical and toxicological properties of the chemical identified in preceding studies permit. Furthermore, mode of administration should be as comparable as possible with other studies. There was a keen sense of awareness about the profound consequences of how a chemical enters into an intact mammalian organism on toxicokinetics and toxicodynamics and the resulting biological consequences. Inappropriate means of administration of a chemical to pregnant test animals could lead to results with questionable relevance for the human population that is potentially at risk when exposed by an entirely different mode and route. The historical database for developmental toxicity hazard evaluations of chemicals in laboratory animals is built on studies conducted mostly by daily oral bolus administrations (gavage) throughout a substantial part of pregnancy. Changes in the duration of chemical administration were introduced in In rats dosing was extended by 4 days from gestation days 6-15 to Therefore, a majority of the studies on record were conducted with the old test guidelines. Gavage administration keeps the daily dose per unit body weight constant as pregnancy progresses into the stage of rat foetogenesis beyond gestation day 15. Only the uterine contents (i.e. foetuses) grow at a dramatic pace during the foetal phase of prenatal development. However, dosing continues based on the total body weight of the dams, including the gravid uterus. 11

15 The scientific rationale for gavage with its associated toxicokinetics and toxicodynamics is difficult to envision in the light of likely real world exposures of humans during normal handling and use of chemicals. Intake by a daily oral bolus, while still endorsed by the OECD 414 test guidelines, is not the anticipated mode of entry of chemicals into the human body. That mode of exposure in animal studies, therefore, appears to be of limited relevance to assess developmental toxicity potential in pregnant women who typically are exposed to much lower daily doses by more protracted means (food and drinking water). The problems created by regulatory classification based on developmental toxicity hazard were brought up repeatedly during the Workshop. Regulatory agency representatives reminded all participants that the Workshop was to focus on chemically induced maternal toxicity and its relationship to developmental toxicity. Regardless, the discussion returned time and again to the problem of overestimating human developmental toxicity potential caused by animal testing hazard-based classification. 3.4 Do further end points improve differentiation between maternal toxicity and manifestations of developmental toxicity? Workshop participants solidly supported acceptance of a recommendation that expanded end point studies might be very useful although there was no unanimous (or close to unanimous) endorsement of that concept. Some attendees remained unconvinced that refinements in detection of the maternal toxic response would help resolving the classification issue since any adverse developmental outcome would still trigger classification as long as the system is hazard based. There was broad support for the recommendation that a search of existing databases and design of study protocols to expand maternal end points concurrent with embryo/foetus end points to assess the developmental consequences could be a useful experimental approach. The exploratory data shown by Dr. Daston suggest that the biochemical factors determined in the acute phase of maternal response, soon after administration of a test chemical, were very informative about the developmental sequelae. Workshop participants who are actively involved in animal studies and experimental work solidly supported the view that selection of 5 or 6 chemicals with equivocal OECD 414 test outcomes as to the origin of adverse developmental effects could become the first step leading to exploring more sensitive end points of maternal toxicity. 12

16 However there was also reluctance to endorse examining expanded end points in order to refine the crude maternal toxicity criteria that have been applied in the past > 20 years in OECD 414 guideline compliant studies. The underlying premise is that the hazard-driven regulatory decision process towards classification and labelling will remain the operating foundation in the future. Even if a maternally mediated mode of action leading to concurrent developmental toxicity seems to occur, it is the developmental toxicity that determines classification. Thus, in the present regulatory framework human exposure conditions during normal handling and use are not given the relevance that they should be given for classification. However, there was curiosity as to how much expression of maternal toxicity might, for example, cause a 10% reduction in foetal body weights. Such information is unavailable at the present time. Systematic experiments have not been performed as to what level of maternal toxicity will produce defined changes in the embryo/foetus. 13

17 4. CONCLUSIONS AND RECOMMENDATIONS 4.1 The current regulatory classifications and their consequences for chemicals that have caused developmental toxicity in OECD 414 testing are causing concern. This is due to the lack of differentiation concerning the origin of adverse developmental effects caused through maternal toxicity (secondary effects) as opposed to those produced by direct events on the embryo/foetus (primary action). The recommendation by a majority of Workshop participants was to move to a process that considers anticipated human exposure during normal handling and use. This would introduce risk characterisation as an important component into the regulation of chemicals in commerce and trade, replacing the current hazard-identification based approach. 4.2 The relevance of route and mode of administration of test chemicals emerged as an important issue for classification of chemicals. Equally weighted were dose-response data as well as toxicokinetics and toxicodynamics, which were recognised as crucial criteria that should be considered in the study design. The scientific evidence that the biological and toxicological effects differ profoundly depending on how the chemical enters into the body can no longer be ignored. Relevant in this context might be recent experimental comparisons in rats of gestational doses delivered by gavage vs. continuous exposure via dosed feed or drinking water. The data indicated that gavage might enhance maternal and perinatal toxicity in some studies ( Comparison of gestational dose [mg/kg] in gavage vs. continuous exposure studies in rats ; S. Parker et al, Abstract #197, Society of Toxicology 43 rd Annual Meeting, March 2004, Baltimore, MD, USA; Toxicol Sci Supplement). If gavage administration is being contemplated, then an adjustment of the daily dose is being proposed, based solely on the dams body weight proper under exclusion of the gravid uterus weight, during the last trimester of rat pregnancy to reduce confounding factors in study data interpretation. The resulting recommendation was that dose-response as well as the rationale for choosing route and mode of administration in OECD 414 testing should be reconsidered when designing such studies. 4.3 The experimental exploration of expanded end points received endorsement from a majority of the participants. Existing databases from previously conducted studies should be scrutinised for toxicity information potentially relevant to the present broadened concerns. New data with refined end points to detect the onset of maternal toxicity might improve the interpretation of the results of such studies compared to the information collected under the present 14

18 OECD 414 test guidelines. An expanded database might provide a scientific basis for industry and regulators in decision-making leading to classification. However, complete consensus was not reached due to the reservations expressed by some of the Workshop participants. There remained doubts that such new data might be constructive as long as classification is a hazard-based process. An experimental way forward to better define maternal and developmental toxicity in OECD 414 studies received broad support. New data with refined end points to detect the onset of maternal toxicity might improve the interpretation of the results of such studies and might provide a scientific basis for industry and regulators in decision-making leading to classification. The recommendation was to evaluate 5 or 6 compounds from among chemicals with equivocal test outcome (primary vs. secondary effects on the embryo/foetus; the so-called grey zone ) for their effects on selected expanded maternal end points (acute phase response, haematology, clinical chemistry) as a means of demonstrating the utility of those added criteria. 4.4 Implementation of the recommendation in item 4.3 would require research funds to conduct the studies. The Long-range Research Initiative (LRI) as a continuation of this LRI sponsored Workshop should deliberate the merits of such experiments. The recommendation is, therefore, that Cefic/LRI should consider funding research to examine the added value of expanded end points in the characterisation of maternal toxicity. 4.5 Once the work proposed in item 4.4 is completed, another recommendation is that the overall outcome should be evaluated in a future LRI-funded workshop. The objective of that meeting would be to determine the research accomplishments. 15

19 APPENDIX 1: WORKSHOP PROGRAMME Tuesday 2 March Registration and coffee Welcome Dr. Mike Gribble Introduction Prof. Geoff Randall Background and Issues of Maternal Toxicity Dr. Frank Welsch Maternal Toxicity: Difficulties in Data Interpretation of Developmental Toxicity Studies and Their Impact on Classification and Labelling Prof. Frank Sullivan Mode and Route of Administration and the Consequences for Metabolism and Kinetics Prof. Ralf Stahlmann Expanded Maternal Endpoints and Their Added Value for Developmental Toxicity Study Interpretation Dr. George Daston Coffee Break Plenary Discussion: What should be the appropriate definition of maternal toxicity? Lunch Breakout Groups: Does pregnancy preclude the use of toxicity data from other studies? What is the impact of mode and route of chemical administration on the manifestations of maternal and developmental toxicity? Do further endpoints improve differentiation between maternal toxicity and manifestations on developmental toxicity? Coffee Break Report and Conclusions from Breakout Groups General Conclusions and Potential Research Directions 16

20 APPENDIX 2: PRESENTATION ABSTRACTS Background and Issues of Maternal Toxicity Frank Welsch Orbitox International Toxicology Consultants Santa Fe, New Mexico 87508, USA Developmental toxicity hazard assessments for chemicals to enter commerce and trade in the EU are presently being conducted and quality assured by regulatory agency test guidelines (e.g., OECD 414). Those guidelines were last updated and revised in An unsettled issue remains the adequate distinction between maternal toxicity caused by high doses of the test chemical, given as stipulated in the test guidelines, and concurrent developmental toxicity. The test guideline criteria defining maternal toxicity remain much cruder than those applied to prenatal development. The OECD 414 guidelines stipulate exposure of pregnant animals to high doses, and maternal toxicity is commonly elicited as a side effect. Concurrently there occur manifestations of developmental toxicity when near term foetuses are examined. The interpretation of the cause of such adverse effects on prenatal development has profound consequences for classification and labelling in commerce and trade. It is known that at times the disruption of maternal homeostasis and well being by chemically induced toxicity may secondarily affect the normal development of the embryo/foetus. However, the prenatally elicited toxicity may also be directly related to administration of the test chemical. The distinction is sometimes equivocal and causes conflicts between scientists who conduct the studies and those in regulatory agencies who are charged with protecting human health and the environment. There are noticeable shortcomings in the presently recorded end points that define maternal toxicity compared to those applied to delineate effects on the products of conception. Exploratory studies are on record that provide additional information to enhance the information provided by more typical end points of maternal toxicity. Such additional data may add sensitivity to detecting the onset of maternal toxicity. Those preliminary results may provide more precise discriminations between chemically induced maternal toxicity causing developmental toxicity indirectly and adverse effects elicited by direct action of the test substance on the embryo/foetus. The introductory presentation to this ECETOC workshop builds on the previously distributed background paper and will set the stage for discussions to review the present state of affairs. The information may lead to research recommendations that may allow less equivocal cause-and-effect associations. A need for additional data may evolve from the workshop presentations and the break-out group as well as joint discussions. Priority recommendations for the most promising research leads might emerge. 17

21 Maternal Toxicity: Difficulties in Data Interpretation of Developmental Toxicity Studies and their Impact on Classification and Labelling Frank M Sullivan Consultant Brighton, BN1 6RE, UK Classification of chemicals for toxicity to reproduction was introduced in the Dangerous Substances Directive 69/549/EEC in To indicate the type of chemical which should be classified the following words were used: Classification of chemicals as toxic to reproduction is intended to be used for chemicals which have an intrinsic or specific property to produce such toxic effects. Chemicals should not be classified as toxic to reproduction where such effects are solely produced as a non-specific secondary consequence of other toxic effects. Chemicals of most concern are those which are toxic to reproduction at exposure levels which do not produce other signs of toxicity. In the past ten years since the introduction of this classification system two rather polarised views have been taken of this wording. The Classification Working Group (CMR WG) of the Commission have tended to focus on the words intrinsic property and since the system is hazard based rather than risk based, have interpreted the Directive as meaning that if any dose of a chemical by any route shows reproductive toxicity, then it should be classified. The Industry scientists have focussed on the words exposure levels which do not produce other signs of toxicity. Because of the serious consequences of classification on subsequent use of chemicals, this polarisation of focus has led to repeated arguments and discussions and recently even to threats of legal action. In this presentation the initial intentions of the classification system will be discussed, along with some factors leading to the present disputes, and some remedies proposed. Analysis of studies on reproductive toxicity both for effects on fertility and on development is complex. Primarily, reproductive toxicity follows the same rules as other types of toxicity and is subject to the same general principles as laid down by Paracelsus in the 16 th Century All substances are poisons; there is none which is not a poison. The right dose differentiates a poison from a remedy. In addition however, some substances have the potential to selectively affect the developing embryo to cause malformations at dose levels below those causing any other adverse effects. Special efforts should be made to identify such substances and to limit exposures of pregnant women, without at the same time classifying a large number of other chemicals, with serious economic consequences, and no benefits to public heath. 18

22 Both Industry and Regulators have tried to adopt a simple single step approach to classification for reproductive toxicity by looking at a single (or several similar) studies and making a decision based on the results. For many of the chemicals now classified the Regulators have said there is clearly evidence of reproductive toxicity therefore it should be classified, and the Industry has responded by saying there is maternal toxicity therefore it should not be classified. This has lead in the past few months to the EU CMR WG asking for a set of rules that can be applied to decision-making on whether the effects are secondary to maternal toxicity or not. In this presentation I will show why this is an impossible request and adoption of a two-step approach is necessary. The first step is a hazard assessment (identification and characterisation), and the second step is a mechanistic analysis to identify whether the observed effects would be likely to occur in humans, at exposure levels of interest. The conduct of developmental toxicity tests used by the chemical industry will be briefly discussed along with the relevance of these with respect to classification. Different aspects of the relationship between maternal toxicity and developmental toxicity will be discussed. On the basis of mechanistic studies, chemicals can be subdivided into various subgroups: One in which the developmental toxicity is not related to maternal toxicity and the chemical should be classified; two classes where the effects are related to maternal toxicity but should not lead to classification; where the effects are seen in the presence of maternal toxicity but cannot be shown to interdependent. In the latter group, the importance of other factors will be discussed which indicate whether classification is justified or not. The relevance of the route of administration and normal handling and use in making the decision will be discussed. The Conclusion of the presentation will be that, considering the objectives of the classification proposals in the Dangerous Substances Directive, a weight of evidence approach is necessary. Decisions based primarily on either any dose by any route or only in the absence of maternal toxicity are inadequate. The mechanism of action, dose, route, class of effect as major or minor, dose-response relationships, presence of a threshold, normal handling and use, consideration of potency cut-offs have to be taken into account. All have to be considered against a background understanding of the consequences of classification. 19

23 Mode and Route of Administration and the Consequences for Metabolism and Kinetics Ralf Stahlmann Charité - Universitätsmedizin Berlin, Germany Institute of Clinical Pharmacology and Toxicology, Department of Toxicology Most drugs as well as other xenobiotics are given orally when tested for their toxicity. The main reason for this approach lies in the fact that human exposure - besides inhalation and perhaps dermal exposure - occurs mainly via the oral route. However, the kinetics of a chemical after its oral ingestion is influenced in many ways, which can lead to major differences in the exposure of an adult or developing organism when compared to the relevant human exposure. It is well known that high oral doses, as often applied during toxicity studies, are only incompletely absorbed from the gastrointestinal tract of small laboratory animals and that the systemic exposure is far lower than expected. In such cases the comparison of the doses applied in animals and man will lead to an underestimation of possible risks. An interesting situation exists with the quinolone moxifloxacin, which shall be mentioned as an example for such a situation. Orally applied doses of moxifloxacin during a teratogenicity test were two orders of magnitude higher than those recommended for therapy (500 mg/kg vs. 5 mg/kg). Nevertheless, the systemic exposure in pregnant rats was considerably lower than the average human exposure (only approx. 25 % based on AUC-values). Interestingly, maternal toxicity was reported to occur under these conditions, which limit the possibility to test the compound at multiple dosing regimens. As a general rule, the following aspects must always be taken into careful consideration when data from a study on the developmental toxicity of a chemical are evaluated: maternal adsorption, distribution, protein binding in plasma and tissue, elimination via the kidney, liver or other routes, first pass effect, placental transfer, distribution and metabolism within the embryonic/foetal compartment and - possibly - aspects of lactational transfer and kinetics of the compound in offspring. Many pharmacokinetic studies show that the kinetics and in particular the metabolism of a compound can be sex dependent and that it differs in pregnancy in comparison to the non-pregnant status as well as in the newborn of juvenile organism in comparison to an adult. Despite the same route of exposure ( orally ), major differences can be found among some compounds if the kinetics of high, toxicologically relevant doses (bolus) are compared with the kinetics of human exposure, which often occurs more protracted. The main determinant of the toxicity of a chemical may be the peak concentration (Cmax), but also the AUC 20

24 values or the length of time above a certain threshold concentration can be crucial for toxic effects to occur. Despite the fact that these relationships are known in principle, they are usually unclear for a chemical when the toxicity test is performed. Major efforts should be made to provide detailed pharmacokinetic data of the compounds tested for developmental toxicity in animal as well as in humans and to select a route and mode of administration which mimics the human exposure most closely to provide a basis for a rational comparison between species. 21

25 Expanded Maternal Endpoints and their Added Value for Developmental Toxicity Study Interpretation George P. Daston Miami Valley Laboratories The Procter & Gamble Company Cincinnati, OH USA It is often suspected that developmental toxicity observed only in conjunction with maternal toxicity is attributable to some aspect of the maternal response; however, the maternal data from standard developmental toxicity study designs is insufficient to fully support such a conclusion. As a consequence, the more conservative assumption that the developmental effects are the direct result of the tested chemical is often taken to be true. As the field of developmental toxicology has progressed, we have discovered a number of potential mechanisms for maternally-mediated developmental toxicity. It is technically feasible to include markers for some of the more common mechanisms into standard protocols as a means of resolving whether the observed developmental effects are secondary to maternal toxicity. A number of maternally-mediated mechanisms that have adverse developmental consequences have been described in the literature. Two of the more common are 1) effects leading to embryonic hypoxia, including decreased uterine blood flow, anaemia, or interference with the oxygen carrying capacity of blood; and 2) effects leading to embryonic zinc deficiency as part of a systemic acute phase response. Examples of agents that induce developmental toxicity by interfering with the mother s ability to deliver oxygen to the conceptus are 1) vasopressin or epinephrine, which restrict uterine blood flow; 2) the non-steroidal antiinflammatory drug diflunisal, which produces haemolytic anaemia in rabbits but not other species; and 3) carbon monoxide, which forms carboxyhaemoglobin and thereby reduces the amount of oxygen that can be carried by the blood. The acute phase reaction is a generalized response to tissue injury or significant inflammation at any site in the body. It is mediated via the release of inflammatory mediators and involves numerous changes in plasma protein profiles as well as the induction of metallothionein (MT), a zinc-binding protein, in the liver. The MT induction is robust and can be sufficiently great in magnitude to evoke a systemic redistribution of zinc, temporarily decreasing the concentration of zinc in the circulation. In the pregnant animal this leaves the embryo transiently zinc deficient, a condition that is developmentally adverse. We have shown that the decreased availability of zinc to the conceptus causes abnormal development, and that a large number of diverse chemicals exert their developmental effects via this maternally-mediated mechanism. 22